Patients with lesions near important functional regions (sensory, motor, or language) pose a particular challenge when planning or executing surgical treatment. Thus, understanding of the individual structural-functional organization and the relationship to the lesion is crucial for surgical planning. Currently functional magnetic resonance imaging (fMRI) has become the most broadly adopted non-invasive method to identify individual functional anatomy to guide neurosurgical decision- making. Conventional fMRI for mapping language function requires that the patient perform precisely timed and cognitively controlled language tasks. Due to the complex and variable organization of language function, to date, there is no consensus on optimal task paradigms that can reliably and completely identify language-sensitive regions. In addition, to improve the predictive value of fMRI, patients usually undergo a panel of tasks, which can be difficult for them to understand and perform. This project aims to develop a novel fMRI paradigm that can effectively map individuals' language function by using a natural viewing condition, where patients only need to watch an entertaining movie clip. Based on recent studies which have found reliable and selective responses in primary sensory-motor and association areas with naturalistic stimuli, the hypothesis is that the temporal responses within functional language areas are also highly synchronized across individuals when they watch the same movie clip consisting of time-varying linguistic content (dialogue, voice over, scene setting), and by detecting which brain areas express similar temporal responses even potentially distorted or reorganized language areas can be detected in neurosurgical patients. We will acquire movie-watching and task-based language fMRIs from healthy subjects and brain tumor patients with lesions in or near language cortex. The objectives of this project are: (1) To investigate BOLD responses from putative language areas of healthy subjects, and extract reliable and synchronized temporal profiles to build language response model, (2) To develop an analysis strategy by comparing different analytic approaches for detecting language areas in individual subjects using the language response model, and (3) To evaluate the effectiveness by comparing with task-based fMRI, and validating against gold-standard invasive language mapping in brain tumor patients. Compared with task-based fMRI, this less-demanding paradigm will be easier for patients to perform (especially those with pre-existing neurological deficits), easier for technologists to administer, and more time- and cost-effective by studying natural language function with only one simple fMRI acquisition. Exploration of this novel paradigm requires development and testing of sophisticated signal processing techniques which will be developed under this proposal.